(Rebroadcast) How to Build a Simple Sports Betting Model | Ep 144

So you're just, there's nothing that anybody who knows how to add and multiply. There's nothing more than that. And I think that the reason I, you know, like that is not going to be enough to win in most things. That is enough to win in some things. I just will be clear. Like if you just construct the problem well enough and are betting into small enough markets, like this could be enough. Hey, what's up everybody? It's GP. I have SP here who just told me that this is going to be the most listened to episode of the, in the history of the podcast. So you know, thank you for joining the millions of people listening to this specific episode. And what do we have in store for them today? Why is this going to be so popular? Well, really setting expectations low to get us started. But no, today we're going to be talking about modeling. And you know, I think a lot of the questions we've gotten recently have been like geared towards interest in getting started in that. I think a lot of questions in the general format of like I'm some type of, you know, top down better or you know, called plusy v betting or whatever these days. But I have, you know, interest in sort of taking the next step. And I think we danced around it a little bit in some of the other episodes. But today I think the goal is to give as like comprehensive of an overview of that as possible. I think definitely we'll start from like square one. But also hopefully we'll talk about some items that are interesting and useful for even people who are like earlier in their modeling journey as well. Yeah. Yeah. And I think I've always, I've done, you know, I've always dipped my toe in audio like modeling stuff. I've always been a little, I've been like, Oh, well, will this work or not? But I think for an intro and just from doing the notes for this podcast, like I'm very, I think that we can do a good job here of getting, getting people at least started on the right track. And then of course, there's going to be some off the tablework as well. So okay, let's, let's, so I think the first question is what is it? What is a sports model? So I'll give you my, my definition of me is I actually had to kind of like, I, I, I, I had to kick her out a couple of different definitions before I felt pretty comfortable. I think it's a mathematical way to predict what might happen in a sporting event. And it's essentially a forecast that takes historical data and uses them to predict future outcomes. Fordict I think is the key word. And I'm sure we'll get into some of that. But I'd love to hear kind of what you think a sports betting model is. Yeah, I think, I think the way we're going to talk about it today, I would, I would agree with that definition. I think like there are sort of two, at least in my head, there's like two classes. There's like models which the idea is to predict the, you know, sort of like from the ground up like what, what the fair price should be. And I think that's mostly like at least how I came in today like talking about this. You know, there's additional types of like hybrid type of model type things where they can be used to price, say, a derivative or, you know, like different models where, you know, because really like you could consider like the types of like angle bettors, model bettors, they're just using one sort of like factor. It's really no different. Yeah. Right. So like, you know, you could have, there's different, I've seen different like models and read about things before where, you know, you can sort of assume market efficiency, like market efficient besides like a couple factors and then sort of like just quantify those one or two or something like that. But I think, you know, in general, the way I think of when I'm, when we're going to talk about like what I think of it's taking some information and turning that into some probability or price. Yeah. And I was exact same thought. Like the thing I was kicking around was like whether to put math, math into my definition or not. And yeah, it's the exact that piece of course you can have a bottom up method. That's like, you know, it doesn't have to be written in code or whatnot, but it could be considered a model. But yeah, I think for how we're both approaching this, like that's kind of the type of model we want to talk about. And I think that's because it's kind of the type that people are interested in learning about, I think is at least from the questions we've got. It's like, you know, how would we build that specific, you know, data science version of a sports betting model? Right. Because that is like a scalable skill that can actually be talked about. Like we can't talk about, we wouldn't be on here talking about, you know, like some sort of, like, right or like it if, like it, basically like an angle type thing because those things are just incredibly fragile, whereas like learning the, the basis of like modeling sort of ground up using data to come up with the prices, you know, I don't think either of you nor I have any problem talking about that because it can be applied in infinite number of ways to infinite number of problems. Exactly. Exactly. Um, okay. So you told me that you had your four stages, you want to kick, kick it off with, with stage one? Yeah. Yeah. Absolutely. So I think, you know, whenever the, you know, these questions get posed, a lot of the time I get like the air of from the person that this is like an overwhelming problem. I don't know how to start. I might not have that much technical background. Like, and then they just, people just don't even start or try it because they, they see like this huge mountain in front of them. Um, and I'm not even here to dispute that like it is a long journey, but like the goal of what I was trying to do when I was putting together my notes for this is to break it into pieces so that people could just take one step, step at a time and feel like they're making progress. So like the first phase for anybody who's like never modeled something before is what I, I called it like the napkin math phase where, um, you, because I think a lot of people get over, like I said, overwhelmed because they hear stuff like statistical terms that they might not know what it means. Like, you know, in, or data science terms that they have no idea what means or distributions and whatnot. They, you know, they just don't have the background or context on that. And what I'd say that like most, most people to start is don't, don't worry about like any of that to start. Like to me, what I would do is, is just, um, try and come up with a price using only, like arithmetic and your domain knowledge. So like, you know, when I was typing on my notes, I, I was used, I was sort of, was using the context of like trying to predict the number of aces, uh, a player's going to have in a tennis match. Okay. This is actually something I've like thought about doing in the past. I don't bet any tennis or anything anymore, but it's something that I think, you know, like, can help give some like actual context to what I'm talking about. So if I was trying to price something like that, what I mean by like the napkin math version of that would be. You would look at, if you're trying to predict how many aces someone's going to have in a match, you're probably going to look at how many aces they average, you know, like in the past five, 10 matches, that would probably be a reasonable starting point. You might also look at like, how many aces there, their opponent allows, right? Like some people are better at returning some or not. And I apologize if, if stuff I say is not right, I don't bet tennis. But this is like how I would go about the problem starting that. I'd look at that. Other things I would look at would be like surface. I imagine matters like, you know, playing on grass, maybe it's faster than clashing play or something, wouldn't probability. Like people who are going to, who are favored should have more aces all else equal than someone not closer matches where people are going to go back and forth. There should be more aces than. So what I would do is like list out, just list out everything you think is important, like in predicting it. And then just try and use like common sense arithmetic, nothing more than adding, multiplying, you know, dividing to come up with what you think like a fairest. So with that, you know, in this example, what that could look like would be, you think, you know, the person serving is more important than the person returning and predicting it. So maybe you put 75% weight on the person serving average aces per game, 25% on the returners average allowed aces per game. You have, you think grass is a 2% bump or something, just vibes or you could Google or something, like, you know, maybe there's like a research paper or something. And then you make some sort of adjustments for like, you know, when they win versus when they lose and like what the probability of that is and stuff like that. So you're just, you're not, there's nothing that anybody who knows how to add and multiply, like there's nothing more than that. I just will be clear, like if you have, like, just construct the problem well, well enough and are betting into small enough markets, like that, this could be enough. But what I, more important what I think is this like destigmatizes it because now once you have this, it's like, okay, now I have the framework of like how I'm thinking about the problem. Now, you know, we can move in the next step, which is, you know, some of the less guessing on, on these, like, what everything's worth and more like the scientific rigor around it, which we could talk about. But like, I just wanted to start here because I think this makes people think, like, actually think about the problem better before just like saying, I'm just going to grab everything and throw it into some machine learning model and hope for the best. I think that's, that's great. That's, that was well said. And I think it's important to get everybody on our side in that this is something that you can actually do. And I just want to add, like, I, SP comes from a technical background, like, I come from like a poker background with no computer science, prior computer science, learning and like everything that I will talk about today is stuff that I learned doing on the side first. And then like, eventually it became, you know, my full-time job and now it's something I focus on a lot, but like, it's completely doable. So I just wanted to also echo that statement as somebody who's kind of been in the shoes of people who are asking these questions of like, yeah, I'm like winning up that, betting in certain, you know, maybe lowering fruit ways. Well, I've gone through that, that process myself and it's, it's 100% doable. So, you know, take what SP said and that's a great first step. And I think that kind of mirrors like some of the very basic first stuff I built, you know, like I think like, like a fairways hit model for golf. I kind of built using a similar structure to what you just outwired, listing everything I thought was important and kind of giving a vibes based approach on like what counted for what? And it's like in tennis, like you have the server and you have the returner and golf you would have like the hitter of the golf ball and then you'd have the course. And like how much does it matter? Like, how are you going to wait like this person's ability to hit fairways and then the course is ability to like give up a fairway hit, right? So it's like, literally you saying that it brought me back to one of the very first exercises I did. So that was very well said. Yeah, I mean, I think that's where basically everybody starts out unless you have like a deep, unless you're coming at it from like a deep data science background or something, right? Like you just are trying to think of the problem intuitively. Like I remember doing something very similar like with different, you know, like college football player stuff for DFS. Like I had no idea in the early days how much, you know, like yard share was worth the versus reception share versus all these things. I just sort of guessed, right? I just would tweak it and guess and say, okay, there's the weight on this is this, the weight on this is this, you know, and over time, like I've hopefully grown and learned like how to maybe more scientifically rigorously wait some of those things and stuff and get a more a better estimate of some of those weights. But it's really the hard work, like 90% of the work is setting up the problem like in a well thought out way. Everything else is really just like refinements for the most part of like estimates and everything. But the problem, like if you don't, if you're not asking the right questions and setting up the problem the right way, nothing else you are going to do is going to, it's not going to like nothing you can do beyond that will make it good, right? Like without this first part of like just the general setup of the problem, like getting precision in your estimates of like how much things is worth. If you're not asking the right questions or framing it upright, it won't matter. So this is why I think, you know, you can win on a lot of things with just this. And I think it's necessary for everything else. Well said. I agree. I agree. I think both we both could say like we've won certain things using just this. Like I certainly can say that. I know you can as well. So like it doesn't have to, it allows you to start and be profitable. And the things that you're going to win are the things that like a fairways hit thing. It's going to be a very small market, you know, you know, it's obviously not going to be one of the one of the main markets that you're going to be able to do this. But like I think we've talked enough on this podcast about how to identify potential areas where you could kind of go in as an individual better without a ton of resources and still win. I think that's like one of our most talked about topics. So like this is the method for that if you want to get into modeling. So I guess we should talk about how to go from that to something more robust. So from, you know, the pen and paper is your next step is your next step, you know, building a database linear aggression like what's what's step two? Yeah. So I what I would I labeled phase two was less less guessing and distilled the actual drivers. So to me, this is like hope the other benefit of hopefully in phase one is you've at least really familiarize yourself with the data. Like it might not be structured in the perfect way, but you know where to go find average aces in the last 10 or you know where to, you know, go find these things. So like the next step is you have all the variables and you have your guesses of what everything is like worth. Here is the point where I think, you know, to do this scientifically as rigorously as possible and you know, try and produce bias and then you know, personal bias and all these things. This is where, you know, introducing some of the data science techniques can be really helpful. And I want to make it extremely clear before I lose people here, like that you don't have to be an expert on these things at all to use them really for the most part, especially some of the more like very basic ones these days. Like you don't need to be like an expert programmer to run a linear regression. Maybe that was true like 30 years ago or 20 years ago. That's not true anymore. And I think anybody can learn that fairly easily. But like what this step is about to me is, yeah, so you have all these factors like, like I said, and you want to get the data in a format where you can actually like test hypotheses and like quantify things. So like I said to me, like I think 99 plus probably percent of problems in sports betting could be solved sufficiently with just very basic regressions like nothing more than that. There are certain, I'll talk a little bit at the end about some some situations, maybe at the, you know, where that's considered other things. But like a basic regression if anybody's not familiar is basically just a way of trying to answer the question of how predictive is this one variable in terms of another variable. So like a simple example, you could like the classic example that's going to be in textbooks is trying to predict ice cream sales as a function of the temperature outside, right? So it's just a, you know, in most programming languages is one line of code. In Excel, you can do these. You don't even need a programming language to run one of these. But it basically, you know, tells you how much something is actually worth. So like in the example, let's just go to the actual example, like in this tennis example, let's say you're trying to predict aces again. And all you put in your regression model is aces per game that the, the, you know, player you're going to bet on has an aces allowed per game by the other person. That would tell you like I think in my first net, I, example, I said way at 75, 25 percent. That might be your hypothesis. But maybe you run that and it says, actually, you want to weigh this at, you know, 63 percent, 37 percent. So you're, you didn't change the question you're asking. You're just refining the, the estimates. And it also like helps, it, it, with the whole context of all the variables, it can answer these questions of what's important, what's not, what things like are driving your projections, what aren't. I think this is how you like refine, refine the estimate. Yeah. I have, I kind of zeroed in on linear regression in, in my notes as well. And I agree with you like a great dataset plus linear regression is pretty, pretty undefeated. But I wanted to talk about like, I wanted to talk about how to structure, like what you actually need to start building a linear regression. So like one thing that SP mentioned is you need the data. So the difference between the difference between the napkin and the regression for the tennis aces is you would have to, no matter if you're running it in our Python or Excel, is you would have to have a database that you're going to let, you know, that, that you're going to let the code run a regression on. So as it will have to be structured out. So you need to figure out, you know, it could be as simple as downloading a CSV from a website from like tennis stats.com or something like oftentimes that's a, that's a viable option. APIs is another, another way to, to get data. But the important part about linear regression that I think makes it good for starting is you're, it's very easy to understand what the output saying. It's like, in the ice cream example, it's like, okay, one degree increase in temperature equals like seven more dollars of ice cream sales. And you know that it's like, you know, doing its best. Basically what it's doing is it's making its best. It's trying to fit a line to the data that minimizes, you know, the distance of each point. Right. So it's doing its best to say it's worth $7. You know, it's not every, you know, every degree increase the data set by $7. But like the takeaway is easy to understand. And that's why I like, you know, I think a lot of people start with it and like it's also very, it's a very powerful tool that you can definitely win with anyway. But I think just getting started doing some basic linear regressions using chat GPT and Python or R or Excel and just kind of being like, okay, that's what this is saying. It'll get your brain going and be like, okay, what other variables can I kind of add to this? That might be important. I think that's when you start to get into the real value add of yourself as a modelers, like what are you going to feed it? And we can talk about like feature engineering as they say, which is just custom, very custom data, you know. But we can talk about that too. But yeah, I just wanted to echo what SP said on linear regression being good, please start. Yeah. The way I, and I think my next phase was sort of some of that where you were going next. But just to maybe dwell on this for one more second. Like the way I think about linear regression is it's like a, it's like a multi tool or something like it's the equivalent of that in the data science world. It's generally flexible and can generally help with a lot of different problems. Like if you're, you know, creative with it and whatnot, you wouldn't want to use, whereas like, you know, there's other tools like a drill or power, power tools or whatever. Those are just more dangerous in the hands of people who don't necessarily like have comfort or know what they're doing. Whereas like linear regression, it's hard to hurt yourself with a multi tool. It's hard to get something like really unreasonable usually, but like get yourself into big problems where you're super over confident or something. It's much harder to do that with linear regression. So that's, yeah, the interpretability, like you had said, of like knowing what you're doing and knowing what you're not doing. You know, we'll talk about some of the other stuff here at the end, but like that's definitely one of the benefits. But I guess moving unless you had anything else on that sort of step. I just want to maybe mention briefly like predictive versus descriptive. We, I think that this will probably come up in in how to evaluate a model a little bit too, but you know, linear regression, I think like is technically descriptive, which basically means like, this is what the line would fit for this set of data. And then, so like, these are, this is how much a temperature, one degree is worth or whatever in this set of data. And then you might be like, okay, I think that that's correct. So in my model, I'm going to use that number to predict ice cream sales or you might be like, well, you know, does that check out if I run this in other countries or does that make sense to me logically or whatever? And you might not always use the exact output of the linear regression to, I guess what we would call like model because that is your strictly trying to be predictive when you, when you're guessing something. Whereas linear regression is like giving you a great descriptive view of the data and often can be flipped and be predictive like as is, but I just want to clarify that. Yeah, it's a good point, like a really good point because you know, one of the most telltale signs that you can know like what someone's model is based on or like how they're thinking about it is if like, if a person is always like showing value on the 56 point dog in college football, they probably have like a regression. They're probably using some type of regression based thing. And that's one of the dangers of regression is regression just by definition is always going to sort of perform better in the center of distributions because there's like, you know, like more data points, like a simple example using the ice cream. And I think this is where you're going is like, you know, you have data points maybe from, I don't know, 30 degrees to hot part of the country called 110 degrees. If there was somehow an 160 degree day, like, I think you're, you're, you're probably predict like a ton of ice cream, but in reality, nobody can leave their house, right? So there's data that, so like that type of thing happens in sports a lot like on the outlier cases. And that's why, you know, regression that that is, I would say it would be like the biggest danger of regression is when you have something that's, and the challenges here is you always have people who say like, everything's like an outlier, like no, this is actually like, so you have a lot of that. But like when you actually do have outlier type situations, regression is generally going to, you know, perform pretty, pretty poorly. Like if you're, you know, any sort of tail thing. So, you know, it is a good point of, of using it. It's, you know, with all these models, I think you mentioned this on the last podcast. They are tools to inform bets. I think how most bettors probably think about it, probably like the top 0.001 percent, you know, like have price, bet price. But I think it's wiser, especially if you're starting off to use it as a tool to inform betting rather than just a tool that tells you strictly what to bet. Right. Like you wouldn't overlay it onto the underdog API and have it guide your best ball practice. Yeah. No, not quite yet. Soon. Soon. Not quite yet. All right. So let's hit up part three. I'm interested in stage three. What do we got? So, for stage three, what I, what I, you know, categorize this one as it was bett test and identify large, what I call it, residuals. That's like a statistical word, maybe a better word. It's just like differences to the market or a difference. You can think of it that differences to the market in case of bets or differences to actuals, depending on like what you're looking at. So once you get, once you sort of like fit a model and you can do this without actually like doing any of the regression part, like let's say you have, you know, you're back of an Appkin math. You come up with prices for aces and you go and you look at the betting board for that day. And you, what I would do is I would try and identify like in total, am I generally aligned with the market first and foremost? And then secondarily, are there specific incidents like where are the cases where like if you're just 5% high across the board, you have some sort of bias in your model. Maybe you're overestimating, you know, the tournaments ungrace. Maybe you're overestimating the impact on grass. Maybe you're, you're not taking in count whether if that matter, matters in tennis or something. So like if you have a systemic bias, that's one of your factors, whether in your regression sort of model or just based on your estimates is off. If you have like specific bias, so like let's say, you know, there's, you're betting the over on every underdog who has, you know, less than a 15% chance of winning or something. What that would immediately tell me is I'm not properly calibrating how the, how like how the win odds are translating to how many aces they're going to get. So let me look at, again, either how I'm fitting that part of my model and the regression or whatever, or let me look at, you know, like my estimates of what these things are worth and making adjustment there. So like that, that's like sort of comparing to the market and identifying and that can be really helpful because I think to your point when we were talking about like fitting the model and everything and finding what features or another word for that is just like what variables like predict things well and what, which ones you're missing. This is how you identify to me, like how things are missing because whenever you build a model, you have what you think is sort of like important and you build in maybe five variables and then you compare to the market and you see, I have bias in these cases. So I need to do something to try and remove that bias or resolve that. So I'm going to add these variables to try and account for it. So like that's this whole step to me, the only other aspect I would say is so you can compare to the market that's very helpful, that's probably the most useful. The other thing I will do before I even bet anything is I'll compare and this is, you know, maybe I'm turning heel and talking about the benefits of back testing here. But this is how this is the point where I think back testing is most helpful. So you fit some model and if you know it's regression based, you can sort of like output what you would think, assuming you're doing everything right and you know, not data snooping and all that, you could fit what you would have made the prices at the time and you can compare those to what actually happened in the game. Like so if you're doing aces, like you could say I projected 10 aces for this guy last year at this match, but he had 35. Yeah, actually, I have a question because are you at this point at this stage, are you just comparing to the, to the prop line and not necessarily like a distribution where you're like betting minus 135, you're like, okay, I would predict like 10 and a half aces at this point in time and either going to look at the market and be like, am I high at 10 and a half aces just like through, just like your mean prediction and then like you're also, I'm sure you're going to talk about like, you know, doing some, some error, you know, analysis on it, but like are at this point, do people need to have built a distribution? Because I do think people sometimes get freaked out by distributions too. Yeah, and so that was, I guess maybe intentional that didn't even mention the distribution stuff yet because like at this point, the way I think about it is, I think most people when they start out, the way they should think about it is they have some model that spits out a single number. It spits out, I think this person's going to have 12 aces. The market is at nine. And then you, you just, just like you can, to me, you know, and we can talk more about the distributional stuff, to me, just like you can go from back of the napkin math to like, into a sort of rudimentary model to more advanced one, you can sort of do back of the napkin math for distributions and like, wear to bet and not. So for example, like in that case, if it was the market's nine and you're 10, maybe you start off saying, I'm only going to bet when I'm X off, right? Like X, you're right, you don't, I guess the reason I don't want to get hung up on, and we can talk about it, but like the distributional piece, like I don't think you necessarily, to bet like any serious way, you definitely need that, right? Because you need to quantify like down, you want to be betting like down to a certain edge factor, right? But early on, I think it's enough to just compare like a mean projection to the market and bet when you're sufficiently off. So at this point, that's, that's definitely what I'm doing. I'm looking at like, I project 12 markets at eight. And this is assuming again, it's like sort of like a normally distributed thing, right? Like if it's not, then that, that sort of goes out the window. But I guess I've been talking a while. Like I'm curious, your thoughts on this or how you do it? No, I think, I think, I think that's super important because as somebody who wasn't from a, like a, like a, like a traditional technical background, I think the distributions were like the least intuitive part. But I did find like if we rewind to the beginning where I talked about like a, a simple fairway model, like that one without ever considering a distribution. I was like, I was like, okay, like I know that, you know, these are the factors. I also don't believe the sportsbooks taking this into account because they didn't adjust, you know, from this to that based on whether or whatever it is. I don't know exactly what the fair, you know, what the probability is, but it's definitely fine to bet. And like that worked. So, you know, I just want to make, I think that that is helpful for, for, especially the people who have asked us questions, like that's a viable way to, to make money. So if you're hung up on like having to have the perfect simulation or worried about, you know, quantifying, especially having some type of, you know, abnormal distribution, right? You know, there's, that's part of it. That's part of betting, but it's not, it's not what you're going to do right away anyway. Like I certainly wasn't trying to like quantify like, ults and, you know, and what not. You know, I was just finding value at finding value at the main and props that were probably not cared about by the sportsbook and like doing exactly what SP is talking about. So, and no, I think about it, like you think about it, but I just want to tell, tell the listeners that like even at this point in time, like there isn't a distribution being used yet. So, you know, don't feel, don't feel too overwhelmed. Right, and it goes back to the, you know, we're trying to build something that helps us make winning bets. We're not trying to predict, we're not trying to make something that is the, you know, God's given truth, right? We're just trying to make something that is going to make us money. So like, to that end, like I'm glad you brought that up because that's, you know, my last phase of phase four. I'm going to mention that a little bit more and for, I make that phase four, eight, I guess, like the distributional part. But yeah, I think, I think just comparing like and identifying where you're sufficiently off market and or where you are, like you weigh over or under predicted what actually happened. Now that's, that's just going to happen. Like you're going to have large residuals in any model. So I'm not telling you to like go and look at those and like again, you projected 10, there's 35. But so there's something wrong with your model. I'm not saying that. What I am saying is, you know, the first thing I'll do, like we'd be even before I bet is I'll look at those, those cases where what I would have predicted is like really far off what happened and just look at like the data I have and the game and my understanding of the sport and say, is there a reason like this was really far off? Like, you know, for example, you know, if you're projecting, let's not do tennis or something, but like if you're projecting passing yards and you projected 500 passing yards and there was 30 and you went and looked at the game and it was like in a hurricane, then maybe that's a thought that like, okay, I need to do something to capture like wind speed better, right? Because it clearly that's like, I'm not capturing that part very well. Because you're basically looking like when I was way off, what was the reason I was off? Sometimes it just random. Like if you were doing this tennis asist thing and a gain or a tennis match went 100 sets and you know, they just kept going back and forth. Like, there's nothing probably you can really predict. Like that's predictive of that. That's just random and that's going to happen. But it can help identify like clear misses. Like if all your misses were on, you know, people who serve the ball really fast, for example, like let's say they're, you know, you don't have serve speed in your model. But all your big misses like are under projecting the people who just serve fast or something or people who are early in their career or something, you're consistently over underestimating, you know, like certain pockets of people, then that's like an indication that you're maybe missing variables that are important to add to the model. So compare to market compare to like what actually happened and that can help you hone in on a better model. You want to talk about variables a little bit because I just want to touch on, you know, because I think when you're talking about sports betting, it's basically like you have data, you use your brain and technology to turn that data into a prediction and then you evaluate it and you make changes or whatever. But when I think about data in sports betting, there's two sets like there's what I would call like raw data. So that could be like play by play data or like basic box score data or whatever. And then there's what people will call, you know, feature engineering, but I don't like that term because I don't think it like makes sense. It's like, I think you said it's just like custom variables, a custom data. It's like using the raw data and turning it into other data that you're going to feed your model. The example could be if you were like doing a passing yards model, you could be like, okay, you know how many yards, you know, somebody passed for last year and you want to think about, okay, will they go over under there passing yards week one of this year? Well you probably want to have like a passing yards per game. So you want to take like their games play and then like use their total passing yards. If you had those two data points, right? And now you've created this is a very basic example, but like something called passing yards per game. And that's like the most basic example, but something something that is basically the crux, I think, of being a good predictive sports betting modeler is taking the raw data and getting as much good raw data as you can and then creating like the custom variables where you need it, that are going to be able to like explain that the model is going to be able to use more efficiently to predict outcomes. Yeah, that was that was square in my last phase, which is iterate and improve. Like the point I had here was like creating your own data. And so like that one of the examples I had was like create your own internal predictive metrics based on other data that you already have. So like you gave a very simple one, I think that's good, but just to give like another flavor of something like like how this can be be thought of is like, you know, EPA in the NFL is like a very big widely used statistic these days. Like expected points basically like, but that didn't always exist, right? Like that was that was someone that made that just like baseball, like baseball, there's all types of statistics which tries to distill, you know, luck from skill and like how good someone actually is. Like the whole reason for doing that is to try and predict the future better, right? Like the reason like, you know, fielding adjusted ERA and all these things exist is because you want to try and remove the noise from like the outcome variable because the things, anytime you can remove noise from stuff and put that into your regression, like you're going to get better predictions, right? So like the idea with a lot of these variables is trying to to put in variables that drive or that capture like the actual drivers of the phenomenon you're trying to capture. So like that's why in like ACEs, I think you probably want to put in something like serve speed, for example, because like that is something just like, you know, golf, like driving distance is relatively predictive, right? Like if someone could like someone can either drive the ball 320 yards or they can't, someone can either serve. God damn it. Just blowing a veggies right here. Someone can either serve really faster. They can't, right? Like, but if you just look at ACEs per game or something, there's a lot of variance there, right? Like there's, were they hitting it right on the line? Like, you know, did they have a good day? But like it's always going to be harder to return the service, someone hitting the ball, 130 miles per hour versus 110. So like that's, that's an example. But like you can, the whole idea of making your own sort of like metrics is in and stuff is to strip out some of like the randomness of these variables and like actually capture the skill. You know, the other things I had for creating your own data that we talked about before and I won't, you know, I don't think this is the podcast. But you know, pulling from less frequent or hard to get places, aggregating from those places or creating or manually tracking your own stuff, right? Like, I think as we talked about tracking who, you know, wins jump balls or something. Like, stuff that's just like not available potentially is another place where you can like feed that into models that takes time right? Because you have to build that up to like get that into your model to actually measure the impact, right? But those can, those can be valuable too. Yeah, I think a good example of that is, is the passing yards like in a hurricane thing you had because you could have the play by play data, but it might not have whether data. I don't know. I don't use the NFL play by play data. So it'll come at me. But let's just assume it doesn't have weather data. Okay. So now you need to go find your weather data, get the time, scan some match up, add that to the database. Oh, wait. All of a sudden like you're missing in a different direction because you're using weather to predict games and domes. And now you don't need to go get a list of stadiums that are domes. And that goes into your, you know, data sets, dome, yes or no, right? And that's kind of the evolution of the data set, which is going to be stuff like that. That's your adding different sources and different, you know, yes, no variables or quantifiable variables like they all coexist in a good database and it'll build up as you gain new insights. Yeah, yeah, exactly. And that's why I just reiterate like the first point of the, to me, the basis is the way you set it up and the questions you answer steps two, three and four, which is again, which I really classified as, you know, refining your estimates, betting, testing and identifying sort of bias and then effectively like iterating and improving off of that. Those last three steps, like are a continual process. Those happen, those will happen as long as you're betting basically for the most part. But then, you know, that first part, the backbone of the model is really, really the most important because the only other things I had that I wanted to mention, you know, with this topic was in the iterated and prove like there's more you can do than just, sometimes you have to do other things beyond just like adding more variables. Sometimes you are truly limited by, you know, some of the aspects of regression, it's rare. Like I said, but like that does happen if something's like really non-linear or there's just a better modeling technique. So I want to even go into that really, but just know there are like other ways to turn, to sort of like quantify things beyond, beyond regression. And they have benefits and downsides, I guess, but you know, oftentimes it's needed if I just say broad leaf, if you're trying to model something that is again, like not linear or has like a lot of interaction effects is where I've personally like needed to use these things the most is interaction non-linear is basically for the most part it for me. I'm curious if you've you know, you've ever ventured into that? Yeah, we do like, we do something for golf that's not, you know, a standard distribution for sure, probably going to leave it at that. But the, I mean, but the thing is like you could get, you could do fun if you had, you know, really good way to predict the players like mean strokes and you use like a standard deviation from data golf. Like I still think that you could do fine. So like this is kind of just like a little extra, but yeah, I think the interact like, I do think that the interaction effects is the key. And we, you know, how much are we going to derail this podcast right now? But it's like, let's say you have some some interaction like like the ice cream, the, let's do the ice cream thing. It's like temperature going up. It's, it's non-linear because at a certain point no one goes outside. So it's kind of like, you know, it's kind of like a traditional, you know, let's just call it a one, you know, one slope line and then it just like falls off a cliff at like 120 degrees or whatever. But what if like there was another interaction effect, which is like temperature and like what time of year it is and then like if it was 100 degrees in January, like nobody would go because they think the apocalypse is coming or something. Like that's like I actually think it's really hard to because of how rare it is a lot of the examples I have for this are like examples off the top of my head that I don't want to just shoot out the dome. But you know, I do think that thing where it's like this might not matter and this might not matter. But if they happen at the same time, they actually matter a lot is this kind of the example. And it's like that is not necessarily going to get captured well by linear regression because linear regression says this matters this much and this matters that much. And it doesn't really care that they have some kind of like a whole is bigger, you know, the bigger than the sum of the parts or whatever. Right. It could be clear like you can get and is where the multi tool thing comes in like you can get really creative with regressions and have add non linear features or transform it to capture that or even add like interaction components. But like when you have it becomes like overwhelming and cumbersome when you have like, you know, if you know two things like clearly interact with each other, like I would just want to make it clear like you can use linear regression and I have used linear regression for that type of stuff. But when you have like three things or they're not clear the interaction or you don't even know really what interacts with each other, that's when it can at least make sense to explore some of these other other methods. And then the last thing that I was going to say that I just mentioned on the distributions is like to me that's that can really be the last step. Like I said, like you said, like I said, I think you can bet without these things for the most part, but to really like turn your mean number or your median number, whatever you're actually projecting into a price, you need to turn that into like a distribution in some way. And that's going to depend a lot on how you're modeling it. Like so if you're using linear regression or whatever type of regression you're using, it has like an underlying distribution that you, you know, is underlying the assumptions and whatnot. But there's different ways to turn that into two prices. And I think that's probably beyond the scope of today. But like that, that would be like the last step so that in the benefit there just to be clear is like you can have precision then because like what you have to do without that is you have to make a judgment of like I was saying, like I project 12, the markets 10, is that a big enough difference such that it's a bet at minus 125? Right. That's a hard question. That's a hard question to answer unless you have an actual distribution. But you still can bet, right? You're probably just going to end up, you just will need to bet and recalibrate, bet and recalibrate of like, okay, two, two ACEs off is I think worth about 10, 10 percent or something, right? Like you just need to recalibrate or calibrate by betting. Yeah. Yeah. You could have, you know, your own push charts for that. And I think that sometimes too, like the problem with the distribution is it could basically do nothing for you if you don't do it properly and it'll take a lot of work. And you might have just been better off like betting the times that you have, you're like two ACEs off market instead of like spending all this time to like price of minus 125 and to a minus 110, when like you might just lose betting those and you might just have one betting like the times your two ACEs off or more, you know, it's like, it's not a guarantee that it's going to be better. And it's, I would say like it's definitely takes some time for you to feel comfortable betting. So when you say, when I say it's minus 125 and the markets minus 110, like if you think about what that really means, like you have to be pretty confident that it's like your minus 125 is pretty good to bet into minus 110, even though if you're thinking of it from like EV threshold, like, you know, it would clear a lot of what people think or like it is an EV threshold using something like odds jam, but like if you're betting bottom up, that's pretty tight window, you know, you have to be pretty confident. Right. You, I've definitely seen cases where people have like a false sense of precision. Like, you know, you see things on Twitter where people post their like projection and then the line and then they're like EV, like like what the EV on the bet is. And I'm like, the projection might be like your mean could be right, but there's no way that's the actual EV. Like, I don't know what distribution you're using to get that, but that is not right. So I mean, honestly, like I think even odd software, it like have to deal with this a little bit if like I know early days like odds jam, I remember tweeting at them and giving them, you know, given a hard time because it's a good way to find it. Because they'd have like, they'd have like, you know, one, one book would have like, you know, points of rebounds assists 16 and a half and another book would be 17 and a half or something. And what they were pricing that at was because that, that's a distributional sort of like problem. Right. This is like the same type of thing is like how much is that worth? It's not like a, for some things, it's, it's more trivial than others, but it's generally not a trivial problem. And yeah, if you're not valuing that appropriately, then you're going to either be like, a lot of times people like, I think are overbending these spots and they, they overvalue what it's worth if they're using the wrong distribution. I think people fall into using one of two things because they're the easiest, the normal distribution or the, um, Poisson. And, you know, I, I've seen a grigis applications about those. And if you're applying those wrong, generally you're going to be like overconfident, you know, especially with the Poisson distribution. So. Well, that, that's a good teaser for our future Poisson episode. When we're really scraping the barrel a full, full episode. It is. I really like this. We can do it. We can do a distribution one sometime. Yeah. I, I, I did see that on the list. I, I think that that would be a good one. But I really like how you set this up and, and how we did this episode. Because hopefully if you're, and, you know, please, you know, to hop in the discord and let us know. But like, hopefully this has made you feel slight, made betting your own numbers, feel slightly more accessible. And if, if we did that, I think that would be, be a win. And I felt like at least we tried. Right. We gave it our all. We gave it our all. Should we go to news? Yeah, I think we got pretty, pretty light news segment overall. Yeah, we're just doing some quick, quick hits on the news, react, react with some really uninformed takes to headlines. And then we'll go over to the questions. So I mean, it's the end is near the end is starting Montana. The great state of Montana, the, you know, a lot of people call it the, the leading state of the US. We all follow it. They have been the first state to ban sweepstakes gambling. It's sad. What do we think? Yeah, my, my, I mean, there's been like a whole host of states who've been doing these things. My initial thoughts these days is like, are the sweepstakes going to just like funnel into the door behind Colchee? And what sport trade I think is trying to do? It seems like my, my take right now is that these, these exchanges particularly are going to be here to stay and they're going to find, they're going to find ways to stay in the game. They're, to me, there just seems like a lot of different loopholes that they would be able to continue to move to. So, you know, if the sweepstakes loophole closes, like, I just, I just don't see them packing it in at this point. Like, I feel like, you know, they would just pivot to either the DFS loophole or the, you know, prediction market loophole or whatever other loophole is, you know, still available to them. I did want to, there was one interesting, I did actually read a little bit about this. I didn't really read about the other one. So, don't worry, those are going to be off the cuff. But this one, they're in the law. They said there is an exception to be made for places that do not allow any real money, like gambling. The actual line was do not allow the use of currency of any kind. So basically saying, like, the sweepstakes that are actually social casinos or whatever are allowed, but if, like, you're doing the fake thing, then they're not allowed. So, they kind of came out and said it. And I was kind of funny. But yeah, I agree with you. I think like, I mean, literally this whole news section is going to hit step by step on each of the ways you can go around the law because the next item is underdog, relaunch their New York DFS. This one happened, I think, just after recorded last week. But they relaunched DFS in New York. I would say the only reason I added this was just in case people were in New York and didn't know. Like, just check it out. They paid a big settlement with the state for, like, operating illegally or whatever the settlement wording was. And now they're back. I don't know. That's pretty wild to me. I don't like what I don't get it. What's New York doing? Like, they weren't allowed, but now they're allowed. I don't know the full story. I believe they like, I was just looking at as you were talking. It seems like they like bought a different, they basically bought their license by acquiring someone and then we're offering like games that were different than like who they acquired. And so they like didn't really pay for their license was the, was the crux of it is what it's sort of seems like to me. Again, we're reading this literally in real time. So well informed. But so now they kind of cleared it up and they're okay to offer it. Yeah. Right. I don't think it was anything like, you know, like bad towards players or anything. No, no, I actually think despite the last podcast, like I think overall underdog, like, is a fair company towards players. I think the like generally, I'm like Dylan was on the podcast a while ago. He's a great guy who works there. But yeah, my, the only grief main group was that I'm not an underdog hater. I just want to make it clear. Just that one decision. No, yeah. No, neither am I. I honestly think they do like a really, a really, really good job of, of marketing and their company is like, well run. Like, I, that's, that's the direction I personally see like a competitor to, to draft Kings and Fandall, like ever coming. I'm not saying it's, it's going to be one of these two. But like, if something were to succeed, my personal thought would be it would come from like the prize picks underdog gamification world. And you see it because, you know, draft Kings released their pick six product, you know, we were talking just a little bit before Fandall seemed at least they're playing release. Something very similar. I think all these companies see that that is such like a, a gateway, like on ramp to customer acquisitions, you know, the DFS thing because you can do it, you know, generally younger. Right. I mean, that's like, but that's like out of the playbook of draft Kings and Fandall is they, got all these people onto their platform. And then there was just the easiest on ramp for sports betting. So I think there's like a huge land grab for, for being like your on ramp to sports betting. And I think underdog prize picks have done a really, really good job in terms of making it super simple, super engaging, very easy to like share and build a community around it and like sweat bets together. Like they've those companies like that have impressed me like so much more than companies than like Caesars, MGM, like just totally uninspired efforts. You know, the underdog, you know, team and whatnot. I mean, he, he did the CEO or founder or whatever, like, I don't know if you know this. He started a company sold it to Fandall. Fandall just like put the caboch on it and basically he started. It was actually my street. I, he may have been involved in that too, but the one I'm thinking I was called draft. Oh, okay. It was like the same idea. It was just like the sort of draft aspects of underdog. And it's clear that they have like a lot more underdog. I'm speaking has like a lot more aspirations because they just came out like a new game. It's like a ladder type game like where you make three picks and like you get higher payouts. Oh, yeah, yeah. People over something. I'm kind of like that. And you know, I've not taken a look at that, but I could see how that's like super appealing to people. So that's, that's the direction. I think like gambling and maybe it's going to be like split into two where you have like this very financialization of gambling. Like if we keep going down this, you know, exchanges and prediction markets and whatnot. But I think most of the money you want and what I mean by that is the recreational money is going to keep pivoting towards like these games, right? And they could ever can make the most fun game is going to be the winner. All right. I think that's a good point. And you know, I think underdog to me is creative willing to listen to people talk to the community. And that's why I was like that I was kind of shocked by it because it just felt like so such a missed up. But I think overall they're certainly a good challenger. We can talk about quickly ESPN bet is launching something like what we talked about. It's kind of like a one stop shop like streaming plus betting. I don't know app that they hinted at. I think it was on the Disney earnings call. I don't know. We talked about that. That was kind of like to us a good kind of their big selling point was ESPN plus is so good for gamblers. These they have like all the off market college basketball right Thursday morning golf like Boondas League of soccer like this is just like like everybody like I'm sure the viewers should have driven like 60% by gambling so it's like please tap into that if you have any shot to survive. Yeah, I think you know the next step I would do if I were them would be lean into like. And this is also one of the things things I think like underdog and price picks have done really well is they've leaned into like content creators who are not just like person getting paid a lot who does not care at all about the product. They've really leaned into like basically they just have a lot both those companies like have a ton of evangelists for their product and like online and I think it's try driven like a lot of engagement. The example that I you know I've mentioned before in this podcast is like I would love to see the amount of traffic that you know book a trend or it's price pick it's got to be insane. Right. So like what I guess to wrap this to the ESPN thing like what I think they should do is try and lean into some of like content creators who are like in the gambling world already and try and get them like integrated with it whether that's like live streaming along or like being the de facto announcer for something. You know what ESPN that John at GoldenPants.com it sweating the golf stream I will do it. I will cross the aisle and stream golf if I can bet one exclamation while I stream. I'm doing and I will do that for you. So just john at GoldenPants.com if you work for ESPN. But yeah I mean you and book it with Trent have a lot of similarities. Well I think that would work well into personality you know level of engagement you know I'm sure that would that would go I think they need like you know like I don't know like people like that like people with where you know fun out of the may be interesting. But what would be interesting is like I think ESPN will probably shot like I think they'll maybe rightly shy away from some of that stuff like I know they've like gone out on a limb with some of the stuff with like I think they have one point they have like bar stool people on and they got a lot of backlash about that. I think that McAfee is still on and I don't think he's still on. So like yeah that's what game that's like who gamblers want to listen to is like you know people like that. Yeah okay I need to reinvent my whole personality then. That's all right you could do a bit I'm sure. Yeah I could be a character I'll play a character for ESPN but 100% 100% that would be kind of fun. Okay the other one this one I thought was interesting and I had actually thought about this before. So the call she and Robin Hood basically hinted at which would make sense if like the sports contracts or whatever we want to call it. We all know what it is. If those come under if those are like considered options or futures or some type of financial instrument then the extrapolation is like theoretically you'd be able to quote unquote trade them in your 401k and call she and Robin Hood basically hinted at that this week that would be wonderful. That would really I would really be fired up about that. Think about tax free sports betting. Are you kidding me? That would be bananas. I would be I mean it's really bad probably for like the overall population but I don't know if something could move the needle for my like personal retirement account like as much as that. I mean what this makes me feel like is it makes me feel like you know someone someone robs a convenience store they steal you know from the convenience store then they get outside and instead of running they're dancing you know like in the street like outside like I know what a dumb thing to say. Like I don't I think they're going to have like I think they have like a good path you know based on everything I see to keep keep this going but like I don't know why you would be I can't imagine this is a good idea to like put these things out in the eat there like I don't I don't understand how this is like going to help your public image like if you think the safe act is onerous like oh my gosh if this thing if 401k vetting is out the whole boy the like deposit limits and stuff like are the you know affordability checks are just going to be out of control like I don't it's me this is like I just we said this it's just so disingenuous to treat this anything other than sports betting and I think some of the things like I've heard these people talk about is like like in this camp talk about I guess is like oh you know like we get better prices for everybody then it like reduces like the harms of like gambling right like the reason gambling so bad is everybody's losing their shirt because the price is right and I could not disagree with that more like if if there was no rake at all like if every bet was even money there would still be people who've ruined their life sports betting correct you know like so the idea that the price is like somehow correlated with like the harms of gambling as people think of it I just wholly disagree with that right like people either can play responsibly and within their limits or they cannot and like a problem gambler is like it's so insane to me because it sort of implies that there's not problem people who ruin their life in the stock market or in crypto or something like like that like it's like they're like no we're gonna you know reduce all the harm of gambling and make it a less scummy industry by by stripping out all the consumers getting like you know raked over the calls and it's just I just couldn't disagree with that more exactly like Robinhood isn't helping anybody's for okay except for no this is the thing it's like except for probably our counterparts who are other you know professional financial traders who are always going to use the 401k to trade tax free and you know and do well and like that's that that's just kind of how it goes but like this easy access to using a 401k to do like the most gambling type of financial products is obviously bad and this is like as much as I would as much as this would be something that I would be like this is going to be helpful like I think I probably would draw a line and say like this is just too far that's a step too far like you know I'll be a sacrifice my person yeah this is this is one of the ones that we just we just give up you know like there's there's a lot of things that would probably benefit you and you and I and you know other others but like not be good for society and I think I generally err on the opposite side right like on these on these issues but this one right you love yeah right it is what them to be lining their pockets with you know this is this one is just it doesn't make it I mean it's just gonna end that it's just gonna end that right like like the I think we the news stories are just writing themselves like this is like what the fake news stories are like like they just don't understand what it is except for like now it's real right you know exactly all right let's talk about people who were in their life gambling I just saw I saw this was interesting it was the Otani betting scandals now getting a movie or his translator I do I do think it it is unfortunate that his name is is obviously attached to this but obviously he's you know what the greatest baseball player is in all time you know and whatever that's how it's why it's a big big story but um yeah I don't know I mean I'm I'm gonna watch it I always get I always like kind of find the gambling movies don't usually to get enough correct um but there are some good ones like I thought rounders and Molly's game generally were good um this will be I think this will be like uh super dramatized and and probably not paint sports betting in the best light but who knows I'll watch it probably yeah I'll watch too I don't know so supposedly like the the bookie behind this all I don't know if this movie I didn't actually see this new story supposedly he is like behind a lot of this and he's writing a book and everything like a bookie so I don't know if that's where it's driven from oh I mean if it's that kind of I think he's actually like on record of saying like he wants to make a Molly's game a movie because I believe he's likely going to jail this guy um at least for a bit um and so if it's if it's from that perspective like I don't maybe there's a bad take I think Molly's game is a reasonably entertaining movie I like Molly's game yeah yeah so like if it's if it has input from the actual okay like if he's involved like you know like it'd be really bad if you know like nobody was involved like it was just like so because you're right you do see some gambling movies that are like did anybody it makes you wonder like when you're watching a movie not about gambling exactly exactly like you just take for granted you're like but yeah you're like wait so I don't know anything about anything because anytime like a movie comes out about gambling or a news story is written like you're just like wow that completely missed anything that makes sense about gambling but um so then you're like wow everything I read is basically bullshit yeah but then like there there are some then there's good gambling stories that come out like the gelco one I thought was really well done so it's like okay it's it's doable Molly's game rounders these are all these are all good things I'm always down for gambling movie there's not enough good ones so let's see let's see um all right let's hit the questions hit the questions yeah all right let's start with this this uh odds jam one let's really like get ourselves roasted by all the 50% use my code odds jammers so it's a good question basically jaw Joshua Royale our best question ask you um he asks how uh how does a company as large as odds jam have staying power in general because presumably sportsbooks can also buy their API package so like odds jam offers many different you know some are basic like here's a bet you should make and others are through optic odds like API packages um that allow you to do a lot of top down ask stuff um so presumably sportsbooks can also buy their API packages just like regular customers and just adjust lines every minute to prevent arbitrage he then posits could it be because odds jam customers help direct them business and that the odds jam customers are somehow an aggregate not profitable betters that's actually interesting I'd love to hear what you're you're take on that is like in aggregate um uh and then he goes on to ask why wouldn't wreck books like mgm just simply copy fandall lines and maybe add a tiny bit more vague or less uh according to their beliefs um to incentivize certain bets and and penalize others as opposed to trying to make their own lines which they have to know are less sharp than the fandall lines so it's there's kind of like this is three questions let's start with the staying power thing um well odds jam obviously we all know sold for a ton of money so odds jam so what odds jam did really really well and so and this is speaking as as somebody who has gppx plus which is like a which is a top down betting product what odds jam does well and why it works is that they have it runs like a mlm so basically like once you burn your accounts you refer people you get part of their odds jam subscription and it allows you to go it incentivize people to go like post come do this because if you have a good top down tool and you find it like let's say sp like your eyes stumbled upon a good top down tool we just didn't think anybody knew about it would we ever tell people to go look at that tool no no exactly so like to have staying power you have to like be able to get new customers in when people burn their accounts or have word of mouth affect so I think that from that standpoint that's how they've had like come customer staying power as to the sports books I mean I do think they do provide services to sports books you know I think they might even sell certain data to sports books that like you know I do think that odds jam and optic odds especially now that they're part of gambling.com group I like they are at the end of the day more in tune they're bigger customers as a sports books so like that's that's just what happens when you're 120 self or 120 million to a gaming affiliate company and that's the reality of it so like do you want to yeah what are your thoughts on this and then I think like talking about the customer aspect of its interest there yeah so my my initial thoughts were when when when odds jam was sold I actually tweeted about this a little bit like that I agree 100% with what you said about like what like odds jam I think it is an incredible marketing company like in like to me that's the different what what differentiates them versus every other sort of top down is they have employed a strategy you can agree with it or disagree with it that really works in terms of getting people to talk about them and get customers on their site so they like really leaned into content and and all these things and like put money behind the referral system like you you said the other thing that I don't think people understand is yeah like who their customers actually are because I also agree that their bread is really buttered from big operator customers more than you know Tommy who's just signed up for his two weeks subscription you know like to really make money in like the sports betting ecosystem like and when I say money I mean like big money like acquisition money you have to get money from the operators it's it's there's just not enough it is my opinion now to me there's just not enough people like that that can that that want these types of things like in the betting ecosystem to like sell for hundreds of millions of dollars like it just it's just not you can have great businesses and make a lot of money on these products and whatnot but like to get acquired for that you have to be making a lot of money from operators because that's where all the money is right like in the ecosystem that's where all the money is going to is the operators so I was like I was running a business and I had aspirations of selling that business for hundreds of millions of dollars all I would be thinking about would be how can I have a business that services the operators because that's where the money is actually like like again when I say money that's where the big money is going to come from um so like I think that's a big component of it to the actual questions then I think most sportsbooks probably just believe like in terms of why they have staying power and sportsbooks buy it like yeah sportsbooks definitely do like license the software I think most sports books just believe and this this hits on a couple layers of the questions I think like because this also hits on like the MGM just copying Fando lines I think sportsbooks either are too lazy or don't do the analysis or probably more likely they believe it's cheaper in their best practice to simply limit customers who you know take advantage of misprice lines and move on rather than invest in technology trading better pricing all these things to me like if I was you know sitting in you know CEO of bed MGM's chair whatever like I would just try and do a cost cost benefit analysis of like okay here's how much we lose every year from people picking off our stale lines here's how much it would take to really like revamp the tech the trading to make sure we're like not getting picked off by these people what is what is worth it and I think probably a lot of companies have done that and determined it's just because the other things that people don't usually think about in this topic is like the tax situation right and like so even like break even customers or even slightly losing customers with like tax taxes and stuff can can be not profitable for the companies so like I think it's probably as simple as that as they just it's easier it's cheaper it's it's better business you know unless you're assuming negligence on their part which I could be true too but I mean these these these people aren't I don't think they're stupid right you know they're they're trying to make the best decision with the resources they have I think I think you're you're exactly right I mean I think like also think about you know I actually so John I've actually Josh's is like a Joshua Al next level programmer I don't think he can even relate to how shitty some of these sportsbooks soft internal softwares are too as well so like even incorporating a simple adjamp API package into their trading software might just be like insurmountable with all of the random things they feel like they have to do on a day to day to like keep the board up as well and obviously at a place like fandall they can do it but like other places yes well a lot of this you know smaller sportsbook they have another thing I don't think a lot of people understand is like it's not like they have a single data feed that generate like populates their whole sportsbook like most of these sportsbooks have like 30 vendors that are all sending their own data feeds and they're all in their own for like it's like a mess that's like you know put together with glue and paper clips or something um you know because different vendors provide like lines for different sports different vendors handle you like your actual like you know like the wallet part of the business and it's just it's just a mess so like again I think it's just like a cost benefit of like is it really worth it? Because frankly like you know these these odds jammers or any other top downers are effectively like doing you know showing them for the price of like relatively low right like to tell them like hey this one's off like I just bet a $3 arb into it like like that that might be cheaper than just hiring like an actual competent traders or someone to revamp the technology so yeah I just wanted to mention all the data because I do think people are think it's like a simple problem but again I think people think these these companies are just like tot like filled with total delts or like people who don't know anything and I generally don't think that's true I think I know they might not be making the best decisions but they're not stupid you know yeah I've talked to traders on you know who work for wreck books who are you know very smart thoughtful people like it's it's yeah there's a lot of competing there's a lot of competing priorities going on that might lead you to be like oh that's maybe irrational but with everything that they have to do to make money maybe it is actually rational I think what you said about the just a note on the the profit better is because I've always I've heard people say this like oh yeah like they'll come in and then like they'll lose they might like are but then they'll like lose on you know they'll like bet part a little keep their mainline like post limits high or so like they should just do that and you'll get these no like I odd gym customers are not profitable customers like you said like a break even a slightly losing customer are still not good customers um and I always think about it from my own point of view like if we have like a month or two of like break even I'm just like holy shit this is a disaster and then I think about like the sportsbooks like also thinking that they're like oh this is it's like no one's no one's having fun we're both losing our costs so it's like because then I always put myself after that and there she is I'm like oh it's expensive to break even in in your actual betting results right I mean they have way higher overhead costs than you do right like even is like a huge disaster for them right so like you know it's interesting to quit to the question of our odds james customers in aggregate not profitable betters if you ask me for in total everybody who's ever signed up for odds jam are they profitable my answer would be definitively no interesting like of all their whole life but that's to me because I'm guessing there is a huge group of people who sign up to odds jam yes yes don't understand adverse selection don't understand variants and don't understand understand understand understand bet sizing and then lose or you know like have a rough start or something you're not the best start quit and then are betting their own stuff yeah so whether you want to count that odds jam or not no no that's okay that counts like this the sum total piano of every better that comes to draft kings or whatever from because of odds yeah I think draft kings is better off for that me personally would be like that's interesting I'm gonna who's actively using odds jam I don't think that's right right but I think I think there's probably a lot of people I mean it's great you you you see people it's like someone goes on five picks in a row and wins and they're like oh this guy is god and lose the ring around so like if I just extrapolate that out to telling someone to bet a two percent edge you know like there's there's people who are gonna just go on a horrible runs on like odds jam and just quit and no I see I see it like GP picks plus it's almost like there's either people who are there for like a month or you've been there for like six months and I know you like you're in Chad and whatever and I've seen you there for six months or like you came for one month and left and like that that does seem to be like the two options for like betting profitable betting tools but then it's like how much are like the six-month there's betting and how much are the one and leave and how bad are the people who go one and leave on odds jam or whatever GP picks or or whatever but if you limit everybody who's been there for three months betting good stuff and then the people who kind of like went in quit and started betting poorly like there's definitely a skewed amount that you could go up compared to go down if you're drafting so I could actually might be coming around to your point of view there that's interesting yeah it's hard to know like you know all these questions it's you know I have no experience on on that side and you know I even have limited conversations about a lot of it so this is all speculation but that'd be my guess and then the last the last one the why I wouldn't wreck looks like MGM just copy fandal lines and add a tiny bit more big or more less big I think we cover that a little bit of like why like it's just a trade-off of like investing in doing that and again you can maybe think that's easy but that when you it's probably not as easy as you or more expensive than you you may think and this is really good question but like oh like this whole thing has been a good discussion but like my thought here on like MGM is or companies like this is these recreational books that have like no interest in getting prices right and everything like they are in I think people confuse what business they are in they are in the business of finding customers who want to bet big who don't care about price or anything that's what they're in the business of so like you know when I think of it I equate them to like you know pretend you had some cake shop they they're they make all sorts of cakes but 90% of their income is from making like custom wedding cakes like these big elaborate custom wedding cakes from like you know these really rich people like that company should probably focus all their time and effort and money and resources on getting more custom wedding cakes they shouldn't try and like save on the margin for their cupcake business which makes like two or three percent so like sports betting is so much of the money comes from so few people you know because you know it's just like the big huge players VIPs that just want to bet that's where like the money comes from I mean you see I can't remember if it was maybe a year ago or so ago like there was a battle in I think a new jersey between draft kings and fanatics about this VIP I think we know VIP like the VIP right like one VIP moving from what you know draft kings to Fandall like literally shifted the market share like in a material way in the state and so like if I'm these companies I frankly do not care about you know these nickel and like these these tiny little better like I am after those people and so again like I don't I really maybe I'm giving them too much credit but I don't think these these companies are stupid this is why I think they spend so much of their time and resources on like marketing and customer acquisition because the benefit of getting like one of those customers to realize that sort of like revenue or profit or anything you have to you have to get so much better at pricing and like get another couple hundred dollars from so many smaller betters right like it's just it's just focusing on the right things honestly that's good advice for everybody and everything you think about just embedding like so much money comes from a few big spots like you should it's hard to correctly divide your attention but like I do think people under under invest in like really smashing good spots and it's the same on the sportsbooks I like that the Jordan question was mostly about starting modeling so I think probably we covered that yeah it was basically like do you have any suggestions or advice for for someone who has a day job and like learning to model and he said he spent a decent amount of time even EV but plus EV betting like I said like you said I think we covered like hopefully you got what you were looking for out of this Jordan the only thing I would I would say would be like I don't I think a lot of people in this case like I wouldn't just replace the top down betting with with like modeling time like I think hopefully we don't do this on this podcast but I think sometimes like the the bottom up can get glamorized it's just being like better more sophisticated or whatever like to me I would keep doing that if you're making money like as long as you possibly can and prioritize that to the what I just said like I would always prioritize where your money is actually like coming from yeah right so like if you're if you're you know tight on time what I always you know think is like and I think I said this before I try and replace my like sweating of games time with something productive like modeling wise and maybe have the like you know habit bundle or something where I can I can have the game on on the side or whatever but also be working on something to try to replace that's not very deep work of you no no I try I you know you got to you got to take what you can get yeah that's that's actually like so important for just work it like take what you can get you don't have to read all the books right so but like that I guess like to the time aspect because we didn't really cover that in the earlier I would maybe do try that like if you're if you are sweating games like try and replace them at that time because that's just like non-productive time that and everybody has their things like I'm not you know everybody needs some time to relax or whatever but then the other things I would say is potentially you know being more effective with your time at work getting some time back there I don't know what your work situation is sometimes that's possible sometimes it's not you could pick up some time there and then lastly I would just try and find a time like whether it's a couple times a week every day you know they're one week same time whatever before work after work I just try and find a time and stick with it because most people have more time than they realize they including myself I just squander a lot of it away because I'm not like I don't I'm not intentional with it all the time but if I am intentional you know I get more done so that would be my advice I've I've one more I'll agree with all those one one more thing to add which actually I find is like a time just like time magic so it's it's like we talked about with the kind of like pair of facts if you consistently work on it you don't have to spend as much total time so what I mean by this is like let's say you did an hour a day even that that's a lot let's say that's the thing you did in the morning for an hour and you just did that every day so that would be my math I think is correct seven hours a week now let's say that instead of that you did like a burst of three hours randomly and then one day you would like do four hours the one hour day is going to yield you more it's gonna move it's gonna move the needle more in seven hours than the three hours one day four hours another day and five days off and I've just found that true with anything like if you want to hack your way around really tight schedule consistency like multiplies the effectiveness of time at least in my interest yeah so like do you have a hard time because I I would say for me I have a harder time like getting started and like like basically like when some in something I can really like like sort of a deep work idea right yeah you don't you find that you you're able like each day is it just like the consistency so you're quicker to get into it because you're doing it every day that's exactly it yeah you start to you you just start to it's like a habit like you're like okay here's my hour boom like I have a well I guess we're on a video called I have a little block that like I just flip over and say good time and it's like okay boom start and if you have it like every day it's like boom I'm gonna start this and then there is a aspect of like at the end sometimes you do want to like say I'm gonna work on this like spend five minutes at the end to say like at the beginning of my hour tomorrow like this is my plan for my hour tomorrow so it helps you kind of just like go on it whenever I've had to like do really big tasks on short time or short on busy schedules that's basically what I've done like an hour a day or whatever but do it every day and then and time it and at the end be like know what I'm gonna do tomorrow so you know so it just rolls over and it's knowing like the first few days but then you just you just start rocking at least for me that that's been effective like everyone has their own and I do agree with you like I would rather work for four hours on something but I find that you don't have as many four hour blocks like as you think and you might end up missing some work here there or whatever I like that I don't know if it was I'm probably miss attributing this I think it was Ernest Hemingway where you know he would write and leave he would he would stop in the middle of a sentence so when he would come back he could start without thinking right he's just finishing the sentence maybe I'll have to try that with like a line of code or something just stop him actually my sister yeah my sister told me about this when I was studying for my my GMAT it's like don't don't finish at the end of a chapter ever yeah you know and I didn't and it was like very helpful so I don't know these are these are all just personal hacks that have helped but I've been there on on very tight schedules I mean I was for a period of time I was doing this and working a investment banking job so you know that was super tight and you know SP is working and doing this as well so you know hopefully these are decent decent steps at least yeah speaking of running out of space time the next question was from Nordicap he's question was working on a large model and running into memory issues to get it running properly any discussion on things like breaking it up a season etc without making the data set too small to get anything out of it I think you had responded that you were you were you know I would probably don't want to get you know too too technical but like how do you usually think of these things well I think one good thing that chat Gbt does very well is it again Nordicap if you've if you've done this you know obviously apologies but just for anyone listening you can give you can give your code to chat Gbt and then you can say hey chat Gbt this is really slow like my computer keeps crashing and I run this could you like refactor this for me or could you just make this use less memory or you know why is it using so much memory and with the the big data sets like oftentimes what helps is you break them into chunks for a lot of like the initial building of the database you know that's just something like chat Gbt will do and then go up them together at the end but like I'm right now basically needing to get a new computer is the other solution is sometimes you just need a new computer or at least get more memory or or whatever it it depends if you've really gone through like the kind of at least asking chat Gbt to make the file size smaller or tell you why it's so large sometimes like there's some obvious reasons like why a file is large and they're not important to the performance of the file and that's something like Gabriel my partner like a really good programmer part of what makes them really really good is making things run faster and use less memory and it's like a massively important skill in programming and one that I don't I don't really have beyond knowing kind of what to ask chat Gbt and how to get it to help me a little bit but I mean I gave the example of that like the pick six contest in that I built it took like a day to run and then Gabriel like made it run in like 15 minutes you know that's like a very important skill and it's it's learned but like there are with AI good starting points for you to ask to speed this up use less memory whatever it's you'll kind of start to know you'll start to kind of understand how to get chat Gbt to take your code and make it you know perform faster use less memory pretty quickly it's one of the best uses of it yeah I just did this recently as well using using some AI tools to to make it run faster I really it's pretty rare for me that that I can't get it to run sufficiently fast just with using those tools it's probably happened before and it maybe you know requires you rethink the problem a little bit you know like if you can determine like there's there's different also like statistical ways to determine do I need like does the model improve or change it all if I use 20 years of data versus 10 you know stuff like that to cut it down because again you're just trying to make sure you're getting like the best conclusions and sometimes you can cut things out but yeah I think in general like using the the AI tools is what I is like what I do I don't I if there was something else you know I could share I would but that's primarily what I do so yeah I think you can't touch on something sometimes you do kind of have to like hack some like proxy things like specifically I'm thinking of times when you'd want to run like a full simulation but it's better to have like a push chart kind of telling you like if you would bet or not it's hard to explain the situation but it's like if you're back testing basically and instead of like running a sim your sim for each event like you kind of have a proxy for how your sim prices things based on the the mean instead of having to run the full sim like there's stuff like that that sometimes at least like we've found we've had to resort to and you know depending on you like like as we said like sometimes there may be a problem that requires like some creativity like that if chat GPT doesn't doesn't tell you doesn't give you a good answer sometimes maybe it's just like not it's not achievable for like a personal computer or something I don't know yeah DFS definitely has those like DFS is not you know like with infinite computer or like computing power a lot of DFS problems are I don't want to call them trivial but they're much easier that like honestly so much of the information is out there of like what you want to do and what you try and need to do and stuff in terms of like contest sims that a lot of the actual like challenge or like where the edges is speed yeah you know like a lot of the best DFS players are like like literally like optimization professors and stuff like so speed is a big big deal there yeah yeah that makes a ton of sense that because the the actual like what you're trying to compute is pretty crazy from like a combination standpoint yeah if you want to make you know 50 if you want to make 50,000 simulations of a 50,000 person field you're you're turning into big numbers very quickly and you that's a case where like probably you need to get creative on how you're doing it rather than just you know generate huge numbers yeah I agree that's a good question that's a very good question I think for this episode it's very very timely okay so we got speaking of models once you crack that code build a successful model we got furious who says he has a working model for tennis head to heads these mainly betting on bet 365 seemingly he's just parlaying he says he's basically just parlaying two head to heads and getting great CLB seemingly maybe moving the line and he's getting bet 365 accounts axed around the four to six thousand in profit mark if I'm reading this this correctly I'm kind of decipher he might be a non-American some of these terms are some of these terms I had to look this up did you did you look this up okay I had to look at these terms so just for the listeners yeah I'll just read it so okay yeah we're not worried he gave the summary of the first paragraph but the second one says due to this gubs as I believe I had pronounced are happening around the four to six k mark how to make it last longer bet on games only 12 hours away rather than one to two days I tried the soup I tried super yanky and hinds as well as almost just doubles crimes any you know you mentioned a primed as well as getting similar results so I'd look these up so this the super yanky and the hinds are like this these different forms of like effectively like types of round robin type parlay things like the hinds is like 57 that's why I think it's called that like it's 57 I think you take like seven bets and it's like seven or whatever like a bunch of doubles some like or in American terms like two leg parlay's through the total amount of three leg parlay's total amount of four leg parlay's all this stuff so it's like a way to to round robin like each yeah successive leg parlay so I think he's I looked it up it's either he's either from I think the UK or Australia he thought he was from the UK but furious you're gonna have to let us know okay yeah yeah hit us in the uh oh yeah I was forgot go and hands out gone click on the button join the discord it's free and you can go to the podcast questions this is these questions so furious let us know in podcast questions if you're from yeah okay as chat you have deep products maybe that's that's what I thought um because it's that gubs is used gubs in the you get it's like getting limited yeah so I I guess my my two cents on this one was if you are if you think you are like the sole generator of the CLV then I would I would bet later for sure um because like you're gonna get you just probably will get less closing line value if you bet like right at post right the line can probably only move so much now like if other people are betting this you know the strategy or whatever these these lines or whatever that's you can't necessarily do that right because you can't wait to the line moves it might not be good anymore right so but like if you are if you believe you are the driver of the CLV I think this is just like broad advice um you probably can wait later and later um because like to me if you're moving the market you want to push that back as far basically until you're no longer moving the market right because you're just gonna get higher and higher limits and whatnot um so that would be the point on CLV I think the other thing I was gonna say is you you mentioned he's getting cut off around four to six K so if I knew a sports book had a limit of of how much you could win before they cut you off because certain sports books like I think they they they cut you off at different like EV thresholds basically like if they think you're you're you're doing some sort of top down they'll cut you off immediately and they may also or in addition to like have like as usually you hit 10k we were you're done like in winning so like you're just done no matter what you're doing or whatever um what I would do in those cases is I want to when I go over that limit of like what I can win I want to go like way over that limit um and so I think that goes to some of the stuff we've talked about about like increasing variance so like let's say you know the you know this example you said bet three six five let's say they're gonna cut you off at I'll call in the pick the middle five K they're gonna cut you off on five K I want to be like losing basically every week but then when I have a winning weight week I want to make like 50k or something um you know so I would lever up my variance like higher just so you're getting more and more out of the account um and so you know I think you're trying some of those things um in terms of like these these brown robins type stuff but like to the extent you can lever that up even more that that's what that's what I would try and do like whenever you know you know a book as a cap you never want to just like hit the cap because then you're making basically the bare minimum you can make on that account um yeah that that's my like bet three sixty five I'm pretty sure does operate like on a win just like if you win this monster just gone that and they're like they are truly one of the most tricky books to infiltrate I'd say um so yeah well as he said it is correct if you're doing like these other brown robin asks it feels like you're finding you know I don't know how many spots you need for a Yankee or Heinz for three or whatever but it seems like you are finding more than just two head to heads um so I would just see if you can figure out like what your what gets you like care more about the to win them out than the risk amount and then also the nice thing about whatever you're doing like the Yankee or the Heinz or whatever is you're gonna have like two leg parles that are very correlated with three leg parles that are very correlated with four leg parles so I mean it's a bankroll question obviously right so like don't go broke just trying to stand to the radar but you know a two win on a kind of like maxed out round robin I'd call it if you're doing like the two's the three's in the four's the two win amount can get very big and it's super correlated so like your bed size becomes very big so if they're allowing like to win on these kind of bigger round robins to be like 30 grand even if it's like each bet would only win you like a thousand or something like that you know if you win like if you really win like it's very likely you'll you'll actually just sweep you know or like come close to sweeping so if you're two win on like a Heinz again I don't know if it's like a four three two or something but like let's say you have a two win on like a four three eight two where you're doing all the fours all the three's all the two's and it's 100k like a lot of your results are gonna be like if it's like a couple thousand to and then you could win 100k or it's you know 10,000 win 100k a lot of your results you'll like lose $7,000 a lot and then you'll win like $50,000 that's a good that's kind of where you want to live if they're allowed now bet 365 because of how they limit like often does try very hard to like cap the upside of tickets to because they because people know this and they try and shoot for like spots that will cash for a large amount and then like you know it becomes a whole game but yeah that like I said that's what I would be aiming for okay we could probably move on to the the last two here okay I can ask this one because I think it's it's directed to you so I'll ask it to you sure coming from a trading background how much more difficult do you believe the financial markets are to win win at than large liquid sports markets sure there's more liquidity the more efficient a market should be in theory but in financial markets there are many parties that are not exclusively focused on maximizing returns some participants are willing to make negative expected value trade in order to hedge a position lock-in-a-fix costs for a commodity etc so he's curious to hear your thoughts on the possibility of market efficiency not increasing in a linear relationship with higher liquidity or limits yeah I think that's true I think that's true like I do think there is a non-linear relationship in financial markets in sports markets I don't know maybe I would say that yes there are obviously like a lot of the commodity trading is done like internally from companies like I know cargill or whatever who are just like hedging commodity positions because if you think of it so they're both optimizing for enterprise value so like the commodity the firm that's producing the commodity like they their enterprise value is greatest if like they don't take on a lot of commodity risk and let's just like assume that to be true so what this question what coolio beans is saying is basically like since firms can maximize their own enterprise value through providing minus ev liquidity to the market does that mean financial markets will could be less efficient than like a high liquidity sports bank market and on top of that sports bank market has a higher rate right so then a financial or higher fees to trade so that's another component of it and I think I actually think that the hardest sports betting markets in terms of in terms of who how many firms can be profitable is a lot lower in sports betting dollar for dollar I'm thinking of like EPL or like soccer like like major soccer leagues like how many how much liquidity is in there and then how many firms can be profitable compared to like I don't know like corn futures or I have no clue what the comparable liquidity has been it's been like 10 years since I've been on a financial trading desk so this is all guesswork but like I do think that it is it is probably a little bit harder to be the hardest hardest sports betting markets than a financial market with the equivalent liquidity and I do think it's partially because of this because like people can participate in the market and make decisions that are profitable for them personally while making minus ev trades in the market like that basically fuels like a lot of the commodity trading but then you're competing with people that are going to be able to realize that liquidity better than you because they invest a lot of money in being the people who are always going to fill those orders and whatever so there's that whole game right so I would say that I would say that my long answer is that there's more there's more money to be made because it's not as adverse of a relationship but there's so much more money put in by firms to making or realizing that that ev that as a solo individual operator it's going to be it might be harder but there is more money to be made and it's probably less efficient yeah yeah this this question for me like in my world this made me think of just like insurance and like how being an actuary you're effectively like you have people who are willingly coming to you to make quote unquote negative ev you know like if you work that insurance company they're coming to you like happily making negative ev bets right quote unquote right but it's it's the case in case of where it's in their best interest to do this and it's in your you know best interest to to do it right so you and I made it got me thinking about like sports too like so much of like peer to peer types of games whether it's exchanges or like if you think about pick six I was thinking about this in this context like you can almost think of it like you know the the the sharp participants or whatever we're effectively writing insurance that the game is suck not exactly right and they're gonna they have to pay out you know like right so like because the game was very high scoring they they've lose their shirt but if the game is terrible like they they collect their money so they it's like I think about that in terms of you know I don't think today right I think the exchanges are just like all sharp like price centric people but as hopefully more recreational liquidity like flows into all of these peer-to-peer games a lot of the what alpha there is going to be is going to be effectively writing the counter side of of people who are not necessarily betting based on financial reasons whether that's you know in sports it's more like entertainment reasons right like you you can collect a premium writing the side that the that's less fun yeah right right you know so and DFS is similar right like um you want to you know collect on like the less popular you you collect when the less popular people do well or whatever so right yeah yeah I think that's a great point it's a great point one end with the why not just why not just flip a coin from my video editor NERK yeah all right go ahead I was gonna read it yeah NERK said had had had a strange thought well a certain circumstance the other day um is guessing on a prop assuming you're thinking it's a coin flip better than putting in a lot of work only to be wrong 49% at the time NERK my man we are we are fighting in these streets you know to get 2% here like you you're here just just wanting to flip a coin you know 2% 2% can feed feed families you can you know put a roof over your head no we we're not flipping coins over here NERK you're better than that that's true he is actually a profitable better so but he's saying you know why not just gamble and someone made a comment on one of my YouTube videos I think it was the the the course one where they're like they're like holy shit this is winning sports winning sounds boring as fuck you know something like along those lines it was like I'd rather just lose or something like that and I was like you know I mean yeah like you know if you're here for the entertainment you would want to flip the coin but the the reality is like like SP said I mean you put in a lot of work and you got a couple percent that's that's the business of the advantage play most of the time so you know those are valuable little percent and we want to flip the we want to find the coin that's 51% us 49% them and then keep flipping it you know so yeah but you know sometimes if anybody has the 51% coin my DMs are yeah yeah sure yeah yeah actually yeah we can we can do this if if NERK's looking for some action well the risk takers podcast will take the 51% side NERK if you're I'll play as long as you want yeah flip the coins and then we can really dig yourself out of the hole and we can make more videos and then we can just do it again now shout out shout out to shout out to NERK my my editor if you're looking for any video content editing he might need the work since he's flipping these 49% coins right now for fun he's gonna live it under a bridge editing your video that's right just flipping hit him in the discord at at NERK 33 and I think his other contact info is in my video descriptions so if you're looking for that but like I said before this time I remember join the discord so you can ask your questions in the podcast questions section or if you have questions or comments about sometimes on the show we're like we we're like I think this is something that might be interesting or any type of you know open-ended questions we air on the show feel free to correct us or ask in the podcast questions as well it's a it's a good interactive experience have people every week engage in the show I I love the questions I love the questions yeah we get we get good turnout so appreciate them all all right well I hope uh I hope people I do hope people got a less scary version of what it's like to start to start modeling hopefully this is a reference type podcast that people could listen to a couple times while while they start and uh if you do have questions about specifically stuff we talked about modeling related like I'd love to get the chat going with with this type of question so we can have a community you know of smart people looking at sports betting modeling ask questions and all help each other out so that's a nice thing about modeling is the skills you learn don't necessarily erode your personal ROI if you share some of them with with other people so it's a great great topic to chat about in the discord so join that goldenpants.com thanks everybody for listening and we will see you all on next episode you